Amazon Invests $50 Billion in AI Infrastructure for US Agencies

Amazon Web Services plans to build specialized AI data centers for US government agencies starting 2026. Investment includes 1.3 gigawatts of computing capacity.

listen Print view
AWS logo on a glass facade

(Image: JHVEPhoto / Shutterstock.com)

4 min. read

Amazon Web Services (AWS) has announced an investment of up to $50 billion in the US to expand AI and supercomputing infrastructure specifically for US government agencies. As announced by Amazon in a statement, construction of data centers that will provide a total of 1.3 gigawatts of AI and supercomputing capacity is scheduled to begin in 2026.

The new infrastructure will be distributed across three specialized AWS government regions: AWS Top Secret for highly classified data, AWS Secret for classified information, and AWS GovCloud (US) for sensitive but unclassified computing tasks. These regions are physically and logically separated from the commercial AWS cloud and meet the specific security requirements of US agencies. The Top Secret regions are air-gapped environments without internet connectivity, while Secret regions are designed for classified information.

The investment aligns with the US government's AI strategy. President Trump recently launched the "$(LB4983764:Genesis Mission)", an initiative to pool AI research resources between government, private industry, and universities. With this announcement, Amazon is positioning itself as a central infrastructure provider for these efforts.

Videos by heise

According to Amazon, federal agencies will have access to the complete portfolio of AI services, including Amazon SageMaker for model training, Amazon Bedrock for deploying AI models and agents, and Amazon's in-house Nova models. Anthropic's Claude models and open-source models are also expected to be available. For hardware, AWS relies on its own Trainium AI chips as well as Nvidia infrastructure.

The expanded computing capacity will enable agencies to handle tasks that previously took weeks. For example, defense agencies can process satellite imagery, sensor data, and historical patterns on a large scale to automatically detect threats and create response plans. Research teams will be able to analyze decades of global security data in real-time. Fragmented data from supply chains, infrastructure, and environmental monitoring can also be integrated into a comprehensive picture.

AWS CEO Matt Garman stated that the investment will "fundamentally transform how federal agencies use supercomputing." He added that it will give agencies expanded access to AI capabilities and remove technological barriers.

Amazon has established a dominant position in US government clouds over more than a decade. In 2011, AWS GovCloud (US-West) was launched as the first cloud infrastructure specifically designed for government requirements. In 2014, the first Top Secret region for classified workloads followed, and in 2017, the Secret region was added, making AWS the first commercial cloud provider accredited for all classification levels.

According to Amazon, more than 11,000 US government agencies currently use AWS services. On the overall global cloud market, AWS holds a market share of 29 percent according to current analysis data, although this trend is declining. Microsoft Azure (20 percent) and Google Cloud (13 percent) are catching up. The massive investment in government cloud could help AWS solidify its leading position in a strategically important segment.

With 1.3 gigawatts of capacity, Amazon is planning one of the largest specialized data center infrastructures for government agencies worldwide. For comparison, a typical large data center has a capacity of 50 to 100 megawatts. The announced capacity therefore corresponds to approximately 13 to 26 large data centers.

(mki)

Don't miss any news – follow us on Facebook, LinkedIn or Mastodon.

This article was originally published in German. It was translated with technical assistance and editorially reviewed before publication.